Mitochondrial DNA Copy Numbers in Pyramidal Neurons Are Decreased

Virginia Commonwealth University VCU Scholars Compass

Neurology Publications Dept. of Neurology

2014 Mitochondrial DNA Copy Numbers in Pyramidal Neurons are Decreased and Mitochondrial Biogenesis Transcriptome Signaling is Disrupted in Alzheimer’s Disease Hippocampi Ann C. Rice Virginia Commonwealth University, [email protected]

Paula M. Keeney Virginia Commonwealth University, [email protected]

Norah K. Algarzae Virginia Commonwealth University

See next page for additional authors

Follow this and additional works at: http://scholarscompass.vcu.edu/neurology_pubs Part of the Neurology Commons

© 2014 IOS Press and the authors. This is the author’s version of a work that was accepted for publication in Journal of Alzheimer's Disease, vol. 40 iss. 2 (2014) 319–330. The final publication is available at http://dx.doi.org/10.3233/ JAD-131715.

Downloaded from http://scholarscompass.vcu.edu/neurology_pubs/6

This Article is brought to you for free and open access by the Dept. of Neurology at VCU Scholars Compass. It has been accepted for inclusion in Neurology Publications by an authorized administrator of VCU Scholars Compass. For more information, please contact [email protected]. Authors Ann C. Rice, Paula M. Keeney, Norah K. Algarzae, Amy C. Ladd, Ravindar R. Thomas, and James P. Bennett

This article is available at VCU Scholars Compass: http://scholarscompass.vcu.edu/neurology_pubs/6

Mitochondrial DNA Copy Numbers in Pyramidal Neurons are Decreased

and Mitochondrial Biogenesis Transcriptome Signaling is Disrupted in

Alzheimer’s Disease Hippocampi

Ann C. Rice1,2*, Paula M. Keeney1, Norah K. Algarzae1, Amy C. Ladd1, Ravindar

R. Thomas1 and James P. Bennett, Jr.1,2,3,4

1Parkinson’s Disease Center, Departments of 2Neurology, 3Psychiatry and

4Physiology and Biophysics

Virginia Commonwealth University

Richmond, Virginia, USA

Running Title: Disrupted Mitobiogenesis Signaling in Alzheimer’s

*Corresponding author: Ann C. Rice, Ph.D. Parkinson’s Disease Center Virginia Commonwealth University PO Box 980599 Richmond VA, 23298-0599 [email protected] ph: 804-828-9664 fax: 804-828-6373

Abstract

Alzheimer’s disease (AD) is the major cause of adult-onset dementia and is characterized in its pre-diagnostic stage by reduced cerebral cortical glucose

metabolism and in later stages by reduced cortical oxygen uptake, implying

reduced mitochondrial respiration. Using quantitative PCR we determined the

mitochondrial DNA (mtDNA) gene copy numbers from multiple groups of 15 or 20

pyramidal neurons, GFAP (+) astrocytes and dentate granule neurons isolated

using laser capture microdissection, and the relative expression of mitochondrial

biogenesis (mitobiogenesis) genes in hippocampi from 10 AD and 9 control

(CTL) cases. AD pyramidal but not dentate granule neurons had significantly

reduced mtDNA copy numbers compared to CTL neurons. Pyramidal neuron

mtDNA copy numbers in CTL, but not AD, positively correlated with cDNA levels

of multiple mitobiogenesis genes. In CTL, but not in AD, hippocampal cDNA

levels of PGC1α were positively correlated with multiple downstream

mitobiogenesis factors. Mitochondrial DNA copy numbers in pyramidal neurons

did not correlate with hippocampal Aβ1-42 levels. After 48 hr exposure of H9

human neural stem cells to the neurotoxic fragment Aβ25-35, mtDNA copy

numbers were not significantly altered. In summary, AD post-mortem

hippocampal pyramidal neurons have reduced mtDNA copy numbers.

Mitochondrial biogenesis pathway signaling relationships are disrupted in AD, but

are mostly preserved in CTL. Our findings implicate complex alterations of

mitochondria-host cell relationships in AD.

Keywords: laser capture microdissection, real-time PCR, neural stem cells,

PGC1 alpha, TFAM

Background

Alzheimer’s disease (AD) is the most common neurodegenerative disease of adults and the major cause of age-related dementia. Memory loss first appears in the disorder known as “mild cognitive impairment” (amnestic MCI) that progresses into AD dementia at a rate of ~15% per year. Biomarkers of MCI include reduced cerebral glucose accumulation with preserved oxygen uptake and increased brain tissue markers of oxidative stress [1, 2]. Progression into

AD dementia is associated with further reductions of cortical glucose accumulation and reduced brain oxygen consumption [1, 2]. These observations suggest that the AD brain may be “starving” metabolically [3].

Studies of post-mortem tissue support impaired bioenergetic metabolism in AD that may play a role in the degenerative loss of hippocampal and cortical neurons. There is loss of activities of decarboxylase TCA enzyme complexes [2,

4-6] and reduced mitochondrial respiration, mitochondrial mass and expression of mitochondrial biogenesis (mitobiogenesis) genes in post-mortem AD brain tissue [5, 7]. Individual AD brain neurons have impaired cytochrome oxidase activity histochemically [6], are populated by deleted mitochondrial DNA (mtDNA) molecules [8] and have reduced expression of nuclear encoded respiratory genes [9] . Loss of cerebral glucose accumulation in MCI and later in AD dementia and direct indicators of impaired mitochondrial function in AD brains could occur from lowered glucose transport, impairment of glycolysis, reduced activities of mitochondrial oxidative decarboxylation in the TCA cycle, impaired mitochondrial respiration, or combinations of these processes.

Mitochondrial respiration involves many proteins encoded both from nuclear and mitochondrial DNA. A complex regulatory system ensures an appropriate coordinated expression from both genomes for both basal respiratory levels or stress-induced upregulation (for review see:[10-13] ). Nuclear

respiratory (transcription) factors 1 and 2 (NRF1 and NRF2) activate transcription

of nuclear-encoded respiratory genes and transcription factor A from

mitochondria (TFAM). TFAM initiates transcription of the mitochondrial-encoded

respiratory genes and has a role in maintaining the mtDNA copy number and

stabilizing the mtDNA. Peroxisome proliferator-activated receptor gamma

coactivator 1α (PGC1α) is a coactivator for many mitobiogenesis-associated

transcription factors including NRF1, NRF2 and estrogen-related receptor α

(ERRα, associated with fatty acid oxidation). PGC1α plays a key role in

responding to environmental changes, through both post-translational

modifications and increased expression, to turn on mitobiogenesis [14]. Although

the exact mechanism of how PGC1α upregulates expression of the nuclear

transcription factors is not clear, increased expression of NRF1 and NRF2 and

subsequent increased expression of TFAM are observed when PGC1α is

overexpressed [15]. Additionally, PGC1α does bind with NRF1, NRF2 and ERRα

and co-activates expression of their target genes resulting in significantly

increased mitobiogenesis resulting in the label of the “master regulator”.

The present study examines in more detail the status of mtDNA and the

expression levels of select mitobiogenesis genes from post-mortem AD and

control (CTL) tissue. We used laser capture microdissection (LCM) to isolate

hippocampal pyramidal neurons, GFAP(+) glia and dentate granule neurons and

quantitative real-time multiplex PCR (qPCR) to demonstrate decreased mtDNA

copy number in pyramidal neurons from AD compared to control (CTL) cases.

We show a loss of correlation of mitobiogenesis factors cDNAs in whole

hippocampal tissue compared to mtDNA copy numbers in AD pyramidal neurons,

which is preserved in CTL cases. In CTL hippocampal tissue PGC1α expression

levels significantly correlate or trend with expression of other mitobiogenesis

genes, but this correlation is lost in AD tissue. Total hippocampal Aβ1-42 peptide

levels from post-mortem tissue did not correlate with mtDNA copy numbers in

any cell type in AD. Additionally, we could not induce the mtDNA copy number

loss by exposing human H9 neural stem cells to the neurotoxic Aβ25-35 fragment.

Methods

Tissue/samples: Human brain tissue was obtained following autopsy and flash

frozen. Hippocampi were obtained from the Brain Resource Facility at the

University of Virginia. Cases consisted of 10 AD by clinical and pathological

diagnosis (mean age 79.9, mean post-mortem interval 6hr, 6 female, 4 male) and

9 CTL clinically considered normal with no pathological abnormalities (mean age

63.2, mean post-mortem interval 8.45hr, 6 female, 3 male). Tissue was embedded in Cryostat mounting media and sliced at 10 µm. Ten to twelve slices were processed for RNA extraction using the miRNeasy kit from Qiagen following the manufacturer’s instructions. RNA integrity was evaluated using the BioRad

Experion capillary electrophoresis system. RNA quality index (RQI) values were

between 6.7 and 9.5 [16]. Representative electropherograms are presented in

supplemental figure 1. The RNA was converted to cDNA using BioRad i-script cDNA synthesis kit following the manufacturer’s instructions. These samples are

referred to as whole tissue cDNA. Approximately 15 slices were processed for

protein isolation for sandwich elisa for Aß1-42 using the kit from Invitrogen and

following the manufacturer’s instructions. Additional slices were melted on

uncoated slides for laser capture microdissection of hippocampal pyramidal

neurons, glia and dentate granule cells.

Laser Capture Microdissection (LCM): Slices were fixed in 70% Ethanol,

hydrated, stained with methylene blue, dehydrated and cleared in xylene for

identifying and capturing hippocampal pyramidal neurons and dentate gyrus

granule cells. Hippocampal glia were identified immunohistochemically using

anti-GFAP (Millipore AB5804) at 1:100 followed by Alexa-488-anti-rabbit at 1:100 for LCM (rt 1hr each). The Arcturus XT system was used for capturing specific

cell types.

Quantitative real-time PCR (qPCR): A BioRad CFX96 thermocycler was used for

all qPCR experiments with BioRad power mix according to manufacturer’s

recommendations for each type of experiment. All samples were analyzed in

triplicate. mtDNA copy number from specific cell types (as done previously by

our group [17]): From pilot studies we determined that capturing 20 cells/cap

(pyramidal neurons and GFAP+ glia) or 15 cells/cap (dentate gyrus granule cells)

and 4 caps/case we could minimize the variability that occurs due to the 10 µm

slice being in different planes of each cell (Supplemental Figure 2). Caps were

7 extracted overnight at 65oC with 50µl tris buffered proteinase K for DNA extraction [17]. The proteinase K was heat inactivated and 200µl TE was added.

Multiplex qPCR was used to determine gene copy number for 4 genes around the mitochondrial genome (12S rRNA, ND2, ND4 and COX III) from LCM isolated cells. The mtDNA copy number was determined/cap of 15 or 20 cells by comparing to a standard curve of circular human mtDNA run on the same plate.

The mtDNA standards were prepared as described previously [18]. Since there was minimal variability between the 4 genes, the average of the 4 genes was used to estimate mtDNA copy number/LCM cap. Primer and probe sequences are listed in the Supplemental Table. Relative gene expression from whole hippocampal tissue: Whole tissue cDNA was analyzed for expression of nuclear encoded mitobiogenesis genes. Relative expression of PGC-1α, NRF1, NRF2,

TFAM and ERRα were determined using Sofast EvaGreen (BioRad) qPCR and primers we designed using Beacon designer software. A panel of 6 reference genes was assessed using Sofast EvaGreen qPCR to determine 3 with the least variability across all cases. GAPDH, 14-3-3z and CYC-1 were selected with

GeNorm analysis in qbase PLUS (Biogazelle) and their geometric means in each cDNA sample were used to normalize relative expression of the mitobiogenesis regulatory genes. All EvaGreen qPCR experiments included a human fetal brain cDNA standard curve for assessing relative expression of each gene. Melt curves were run with each qPCR to ensure only a single species was being amplified. Specific primer sequences used are in the Supplemental Table.

Aß1-42 elisa: Slices were dissolved in 5M guanidine, 50mM tris (pH 8.0) plus

protease inhibitor cocktail (Calbiochem, set III). Protein concentrations were

determined using the Pierce BCA protein assay with BSA as the standard.

Samples were diluted with 0.1M phosphate (pH7.4) to achieve 1mg/ml.

Sandwich elisas were analyzed using 50µl of each sample in duplicate following the manufacturer’s protocol (Invitrogen). Most AD cases were much higher than

the CTLs, so they were re-analyzed shortening the time of the horseradish

peroxidase visualization step so the high end of the standard curve and those

samples had absorbance values within the detection limits of the

spectrophotometer.

Cell Culture H9 neural stem cells: Human Neural Stem Cells (H9-derived) (H9

NSCs) were purchased from Gibco (Life Technologies) and cultured according to

the supplier’s instructions. Briefly, cells were grown in Knock Out DMEM/F12

with 2 mM GlutaMax-I, 20 ng/ml basic Fibroblast Growth Factor, 20 ng/ml

Epidermal Growth Factor, and 2% StemPro Neural Supplement in CellStart

coated vessels. All medium components were purchased from Life

Technologies. Medium was changed every 2 to 3 days and cells were kept at

37° C in a humidified CO2 incubator. H9 NSC cultures were maintained

between 50 and 90% confluence. Aβ25-35 and Aβ35-25 (Bachem, Torrance, CA)

were diluted to 2 mM in water and aliquots stored frozen. 72 hours before

treatment aliquots were placed at 37°C to aggregate [19]. 35 mm dishes of 90% confluent H9 NSCs were treated with 10 uM aggregated Aβ25-35 or Aβ35-25 in

growth medium. After 48 hours cells were lifted with Accutase, washed, and

9 each pellet sonicated in 350 ul RLT Plus (Qiagen) with 1% 2-mercaptoethanol.

DNA and RNA were isolated from triplicate samples for each treatment using an

AllPrep DNA/RNA Mini Kit (Qiagen). Genomic DNA was quantified using a

NanoDrop 2000c.

Statistical Analysis: All analyses were performed using the statistical software by

GraphPad Prism. Frequency distribution of mtDNA copy numbers determined for

each LCM cap from the qPCR experiments were divided by the average mtDNA

copy number/cap of the control cases for each cell type. A frequency distribution

of each cap as a percent of CTL was analyzed using Kolmogorov-Smirnov test

for comparing cumulative distributions after outliers (±2xSD) were excluded.

Between AD and CTL for each cell type a Mann Whitney non-parametric t-test

was used to obtain p-values. P<0.0167 was considered statistically significant.

Correlation analyses were performed between relative expression of the

mitobiogenesis genes as determined by obtaining the relative expression of each

gene normalized to the average of the 3 reference genes used for each case. A

non-parametric Kruskal-Wallis analysis was performed to assess the correlation

between PGC1α and the other mitobiogenesis genes from AD vs CTL cases,

between the average mtDNA copy number/case from hippocampal pyramidal

neurons and the normalized relative expression of the mitobiogenesis genes and

between the Aβ42 levels/case and the average mtDNA copy numbers/case.

Results

10 mtDNA gene content is reduced in AD hippocampal pyramidal neurons. To determine if the mtDNA copy number was reduced in hippocampal pyramidal neurons, GFAP(+) astrocytes or dentate gyrus granule cells in AD cases compared to CTL cases, we isolated specific cell types using LCM and analyzed for mtDNA gene copy number for 4 genes around the mitochondrial genome

(12S rRNA, ND2, ND4, COIII; see Methods). Ratios of mtDNA protein coding genes to 12S rRNA were all ~1.0 (Supplemental Figure 3), did not suggest any significant level of mtDNA deletions in our populations and demonstrated that we were only assessing DNA (not RNA) from our extraction. The mtDNA copy number reported is the average copy number of these 4 mitochondrial genes/LCM cap. Mitochondrial DNA copy number values were divided by the mean of the CTL cases for each cell type and presented as a percentage of mean CTL values. Figure 1 depicts a histogram of the fraction frequency (in bins of 20) of mtDNA copy number for each LCM cap as the percentage of the mean of the CTL group for each cell type. The AD pyramidal neurons (Fig. 1A) are shifted to a larger proportion being less than 100% compared to CTL (Fig. 1B) indicating significantly decreased mtDNA copy numbers in hippocampal pyramidal neurons from AD cases (p=0.0086, non-parametric Kolmogorov-

Smirnov test). Mitochondrial DNA copy numbers in glia from AD (Fig. 1C) were not statistically significantly different from CTL (Fig. 1D), although there was a trend towards significance (p=0.0899, non-parametric Kolmogorov-Smirnov test).

Mitochondrial DNA copy numbers in dentate granule cells were not significantly different in AD compared to CTL (Fig. 1 E,F). A t-test of the raw values of the

11 mtDNA copy numbers from hippocampal pyramidal neurons also indicated a statistically significant decrease of 17% (p=0.0475) in AD compared to CTL (data not shown).

mtDNA copy number in CTL but not in AD pyramidal neurons correlates

with cDNA levels of mitobiogenesis factors NRF2, ERRα, NRF1 and TFAM.

To determine if the decreased mtDNA copy numbers in AD pyramidal neurons

were related to the relative expression of the mitobiogenesis genes, we extracted

total RNA from frozen sections of the same tissue block used for the specific cell

isolation and generated cDNA using random hexamer priming. The relative

expression of each gene was normalized to reference genes with the least

variability across all samples (See Methods). Figure 2 shows the relatively linear

relationships among levels of pyramidal neuron mtDNA gene copy numbers

(average/case) and the reference-gene normalized expressions of the

mitobiogenesis factors NRF2, NRF1, ERRα and TFAM in CTL (Fig. 2B, D, F, H),

but not AD pyramidal neurons (Fig. 2 A, C, E, G). PGC1α showed a trend toward

a linear correlation, but did not reach statistical significance (Supplemental Figure

4). These experiments indicate that in post-mortem hippocampal tissue from

CTL cases the relationship between mitobiogenesis factors and mtDNA copy

number in pyramidal neurons is preserved, whereas this relationship is lost in AD

cases.

Mitobiogenesis signaling relationships are preserved in CTL but not AD hippocampi. To determine if the relationship between PGC1α, the “master” upstream co-activator of mitobiogenesis signaling factors, and expressions of

NRF1, NRF2, TFAM and ERRα were preserved in each case, we evaluated the correlation between the relative expression of PGC1α and each of the other genes/case. Figure 3 depicts the significant correlation between PGC 1α and

NRF2 and ERRα in CTL cases (Fig. 3B & H,p=0.017 and 0.021, respectively), but not in AD cases (Fig.3A & G;). NRF1 also showed a positive trend in CTL cases (Fig. 3D), but did not reach statistical significance (p=0.085). In both AD and CTL, TFAM trended toward a correlation with PGC1α (p≈0.2). These experiments indicate the correlation between the master regulator PGC 1α and the downstream mitobiogenesis factors in post-mortem tissue is preserved in

CTL cases, but lost in AD cases.

mtDNA copy number does not correlate to Aβ(1-42) levels. To determine if

Aβ1-42 levels were related to the decreased mtDNA copy numbers in AD pyramidal neurons we prepared whole tissue homogenates from the same tissue blocks used to isolate specific cell types and analyzed for Aβ1-42 levels using a sandwhich elisa. Figure 4 shows the lack of correlation of mtDNA gene levels in isolated pyramidal neurons, GFAP (+) astrocytes and dentate gyrus granule cells compared to Aβ1-42 levels in AD. Interestingly, CTL hippocampal Aβ1-42 levels positively correlated with the pyramidal neuron mtDNA copy number from each case (p=0.0038; two cases had moderate Aβ1-42 levels). This correlation must be

13 viewed in the context that it is being driven by the two CTL cases with moderately elevated Aβ1-42 levels. Additional correlations of Aβ1-42 levels with mitobiogenesis

gene expression levels were also not significant in AD cases. However CTL

cases showed significant correlations with Aβ1-42 and NRF1 and ERRα and close

to significance (p=0.055) for NRF2 (Supplemental Figure 5). The lack of

correlations in the AD cases may be skewed by the one AD case with low Aβ1-42

levels.

Exposure of human H9 neural precursor cells to neurotoxic Aβ25-35 does not

reduce mtDNA gene copy number. To determine if we could induce the

decreased mtDNA copy number we observed in post-mortem AD tissue, we

attempted to replicate in a non-tumor, human neural stem cell model the loss of

mtDNA copy number observed in LCM-isolated AD hippocampal pyramidal

neurons. Human H9 neural stem cells (derived from a hESC line) were exposed

to 10 µM neurotoxic Aβ25-35 fragment for 48 hrs and ~50% loss of cells was

observed. Figure 5 shows that this treatment did not significantly alter mtDNA

copy numbers when compared to vehicle treated cells or cells treated with the

inverse sequence Aβ35-25.

Discussion

In our postmortem AD hippocampi we found that mtDNA gene copy

numbers were significantly reduced in surviving pyramidal neurons compared to

identical cell types in CTL hippocampi. This is consistent with the observation of

14 reduced mtDNA copy numbers in whole hippocampi [7] and cortex [20] from AD

cases. However, Hirai, et al., [21] using in situ hybridization detected increased

mtDNA in hippocampal pyramidal neurons from AD cases compared to CTL;

from EM photos they determined many of these were in lipofuscin vacuoles and

are likely from degrading mitochondria. Additionally, they did not detect mtDNA

differences in the glia or granule cells by in situ hybridization, which is consistent

with our observations in those cell types. They also comment that using PCR

techniques, they did not detect an overall difference in mtDNA in AD compared to

control tissue. Our cells were isolated by laser capture and only intact cells were

selected, which could explain the difference between our study and their study.

Another factor that could be contributing to the decreased mtDNA copy number

in AD in our study is that the control cases were younger (~16yr) than the AD

cases. Our group has previously demonstrated dramatically decreased (~90%

loss) mtDNA copy numbers in whole brain tissue from 21 month old mice

compared to 5 month old mice [22]. Since our samples were isolated from

selected intact cells, we would not expect as dramatic a difference due to age as

is observed in whole tissue studies. A comparison of mtDNA copy number/LCM

cap or mitobiogenesis genes versus age revealed no correlation in either AD

cases or CTL cases in our cohort of samples (Supplemental Figure 6).

Additionally, unknown factors about the individual’s pre-mortem status, such as

medications they may have been on, could also have affected the mtDNA copy

number values we determined.

Recently, cell-free mtDNA copy numbers in cerebral spinal fluid (CSF) from AD subjects, asymptomatic at risk subjects and pre-symptomatic subjects

with a pathogenic PSEN1 mutation have been shown to be markedly reduced

compared to age matched control subjects, non-pathogenic PSEN1 mutation

family members, or individuals with a different dementia pathology (FTLD),

These findings indicate that the reduced mtDNA copy numbers both occur prior

to clinical symptoms appearing in AD and appear to be selective for AD dementia

[23]. The same authors reported that levels of mtDNA were reduced in cortical

neurons derived from a transgenic (APP/PS1) mouse model of familial AD. They

conclude that reduced CSF mtDNA levels appear to be a robust biomarker for

pre-AD dementia. They also suggest that the significant loss of mtDNA copy

numbers we observed in hippocampal pyramidal neurons and nearly significant

loss of mtDNA in GFAP(+) astrocytes have clinical correlates in CSF and may

represent early pathogenic events in AD.

Previous studies from our group have demonstrated decreased mRNA

expression of mitochondrial encoded genes in cortical AD tissue homogenates

from the same tissue sources used in the present study [5]. Decreased mRNA

expression of many of the mitobiogenesis genes (PGC1α, NRF1, TFAM, POLγ

and PPRC1) were also detected in the cortical AD tissue homogenates

compared to CTL [5]. In hippocampal tissue homogenates from AD cases,

Sheng, et al., [7] observed decreased mitobiogenesis protein levels compared to

control tissue (PGC1α, NRF1, NRF2α, NRF2β and TFAM). Subsequently, when

they stably knocked down PGC1α in a cell culture line, they observed the

16 expected decreases in protein levels as well as reduced mtDNA levels relative to nuclear DNA levels by about 50% [7], supporting the regulatory hierarchy of

PGC1α upregulating NRF1 and NRF2 leading to increased levels of TFAM and

subsequently increased mtDNA levels. Our decrease in mtDNA copy number in

AD cases was not quantitated the same way as Sheng, et al. [7] and our

decreased AD pyramidal neuron mtDNA copy number was not as dramatic as in

their cell culture. Our results, obtained from laser-captured, postmortem AD

pyramidal neurons complement those Sheng, et al., [7] obtained in cell culture.

We also observed a trend towards decreased mtDNA copy number in

laser-captured GFAP(+) glia. Even though it only approached statistical

significance, it does demonstrate that the glia, while not thought to be affected in

AD, may in fact also be metabolically impaired and contribute to the deficits other

groups observe in whole tissue homogenates. Additionally, we observed no

difference in mtDNA copy number in dentate gyrus granule cells, suggesting that

they may not be bioenergetically altered in AD, and adding cell specificity to our

findings in pyramidal neurons. However, we cannot confirm bioenergetic status

in individual post-mortem cells isolated by LCM. Additionally, since we only

selected healthy appearing cells, we cannot speculate what may be occurring in

unhealthy or dying cells or what may have led to the death of cells no longer

available for sampling.

We uncovered a more generalized defect in AD hippocampi, that of a

dysregulation of mitochondrial biogenesis signaling. Using pyramidal neuron

mtDNA copy number as an indicator of mitobiogenesis, we found in CTL

17 hippocampi generally linear relationships between expression of upstream mitobiogenesis genes (NRF2, NRF1, TFAM and ERRα) and mtDNA gene copy numbers. The relationship between CTL pyramidal neuron mtDNA copy numbers and PGC1α expression levels trended towards significance. Since

PGC1α activity can also be modulated through post-translational modification, its mRNA levels represent only one level of regulation (transcriptional) that may have limited correlation with the expression levels of the other mitobiogenesis genes or with the mtDNA copy numbers. These data indicate that in control tissue from autopsy cases the hierarchy of mitobiogenesis regulation appears to be intact. However, in AD tissue this correlation was missing, supporting the idea of impaired mitobiogenesis in AD as purported by others [5, 7]. Microarray analysis of LCM isolated hippocampal pyramidal neurons from AD cases has shown decreased expression of nuclear encoded respiratory genes, which is also

consistent with impaired mitobiogenesis [9].

We also found in CTL, but not AD, hippocampal cDNA significant linear

relationships among expression of the “master” mitobiogenesis regulator PGC-1α

and multiple downstream mitobiogenesis factors it co-activates (NRF2 and

ERRα) and a trend towards significance in others (NRF1) [11, 12]. Due to the

heterogeneity of human samples, had our n values been larger, we may have

achieved better correlations between all the mitobiogenesis genes expression

levels. These studies also support an intact mitobiogenesis hierarchy in CTL

autopsy tissue that is lost in AD tissue. Future studies will examine this

relationship specifically in pyramidal neurons.

A significant correlation between PGC1α and two key metabolism related proteins (pyruvate dehydrogenase A1 and complex III subunit UQCRH) has been demonstrated in skeletal muscle biopsies from diabetics (type II), and those with and without a family history of diabetes mellitus [24]. In a model of muscle disuse atrophy, Cassano, et al., [25] demonstrated the expression level of

PGC1α significantly correlated with both NRF1 and TFAM levels in the muscle

atrophy acetyl-L-carnitine–treated animals, but only with TFAM in the non-treated

control group. Thus, mitobiogenesis factors seem to maintain their hierarchal

correlation in other diseased tissues, while in AD hippocampal tissue it does not.

The origins of this substantial dysregulation of mitobiogenesis signaling in

AD are not clear from our experiments, likely are multifactorial and may have

different molecular origins among the AD cases. PGC-1α mRNA and protein

have been reported to be decreased in AD brain, suggesting that dysfunction of

mitobiogenesis may be the origin of mitochondrial dysfunction observed by

several groups [5, 7]. Speculation as to potential mechanisms of mitobiogenesis

disruption would incorporate many potential etiologies and is beyond the scope

of this discussion. Future studies will investigate the potential role epigenetic

regulation may be playing.

While some studies correlate Aβ1-42 levels with AD pathology, we found no correlation of Aβ1-42 levels to the mtDNA copy number reductions in hippocampal

pyramidal neurons, glia or dentate granule cells from AD cases, which is similar

to what our group found previously in cortical tissue homogenates [5]. Our

limited experiments with exposure of H9 neural stem cells to Aβ25-35 peptide do

19 not support a primary role for Aβ peptide mediating loss of mtDNA gene copy

numbers acutely. However, the limitations of this experiment include using a

more neurotoxic fragment and not Aβ1-42 peptide itself, and short incubation times

that were necessary given the degree of cell death produced in 48 hrs by Aβ25-35 peptide. Longer term exposure of human neurons to Aβ1-42 peptide, particularly

oligomers at low concentrations, will help test the hypothesis that Aβ1-42 peptide

is a pathogenic factor in loss of mtDNA genes. Future studies will examine this

relationship in these human neural stem cells after they have been differentiated

into neurons.

Our findings show that maintenance of normal mtDNA gene levels within

individual pyramidal neurons or astrocytes is defective in AD, and that

mitobiogenesis signaling at the transcriptome level is preserved in CTL but not

AD hippocampi. Whether mitobiogenesis signaling is also disrupted in

vulnerable AD pyramidal neurons remains to be studied by LCM approaches.

Because post-mortem studies can only find correlations and not test causal

mechanisms, our findings support but do not prove the concept that abnormal

mitobiogenesis signaling and disrupted downstream responses contribute to

metabolic deficiencies observed in AD and may be primarily pathogenic.

Furthermore, our cohort of samples is small and displays considerable variability,

therefore we caution against generalizing our results to the sizable population of

AD sufferers.

Acknowledgements: This research was supported by the Parkinson’s and

Movement Disorders Center at Virginia Commonwealth University through the

Medical College of Virginia Foundation.

References

[1] Mosconi L, Mistur R, Switalski R, Tsui WH, Glodzik L, Li Y, Pirraglia E, De Santi S, Reisberg B, Wisniewski T, de Leon MJ (2009) FDG-PET changes in brain glucose metabolism from normal cognition to pathologically verified Alzheimer's disease. Eur J Nucl Med Mol Imaging 36, 811-822. [2] Gibson GE, Shi Q (2010) A mitocentric view of Alzheimer's disease suggests multi-faceted treatments. J Alzheimers Dis 20 Suppl 2, S591-607. [3] Mamelak M (2012) Sporadic Alzheimer's disease: the starving brain. J Alzheimers Dis 31, 459-474. [4] Manczak M, Park BS, Jung Y, Reddy PH (2004) Differential expression of oxidative phosphorylation genes in patients with Alzheimer's disease: implications for early mitochondrial dysfunction and oxidative damage. Neuromolecular Med 5, 147-162. [5] Young-Collier KJ, McArdle M, Bennett JP (2012) The dying of the light: mitochondrial failure in Alzheimer's disease. J Alzheimers Dis 28, 771-781. [6] Cottrell DA, Borthwick GM, Johnson MA, Ince PG, Turnbull DM (2002) The role of cytochrome c oxidase deficient hippocampal neurones in Alzheimer's disease. Neuropathol Appl Neurobiol 28, 390-396. [7] Sheng B, Wang X, Su B, Lee HG, Casadesus G, Perry G, Zhu X (2012) Impaired mitochondrial biogenesis contributes to mitochondrial dysfunction in Alzheimer's disease. J Neurochem 120, 419-429. [8] Krishnan KJ, Ratnaike TE, De Gruyter HL, Jaros E, Turnbull DM (2012) Mitochondrial DNA deletions cause the biochemical defect observed in Alzheimer's disease. Neurobiol Aging 33, 2210-2214. [9] Liang WS, Reiman EM, Valla J, Dunckley T, Beach TG, Grover A, Niedzielko TL, Schneider LE, Mastroeni D, Caselli R, Kukull W, Morris JC, Hulette CM, Schmechel D, Rogers J, Stephan DA (2008) Alzheimer's disease is associated with reduced expression of energy metabolism genes in posterior cingulate neurons. Proc Natl Acad Sci U S A 105, 4441-4446. [10] Scarpulla RC (2008) Transcriptional paradigms in mammalian mitochondrial biogenesis and function. Physiol Rev 88, 611-638. [11] Scarpulla RC (2011) Metabolic control of mitochondrial biogenesis through the PGC-1 family regulatory network. Biochim Biophys Acta 1813, 1269- 1278. [12] Scarpulla RC, Vega RB, Kelly DP (2012) Transcriptional integration of mitochondrial biogenesis. Trends Endocrinol Metab 23, 459-466. [13] Hock MB, Kralli A (2009) Transcriptional control of mitochondrial biogenesis and function. Annu Rev Physiol 71, 177-203. [14] Wu Z, Puigserver P, Andersson U, Zhang C, Adelmant G, Mootha V, Troy A, Cinti S, Lowell B, Scarpulla RC, Spiegelman BM (1999) Mechanisms controlling mitochondrial biogenesis and respiration through the thermogenic coactivator PGC-1. Cell 98, 115-124. [15] Wareski P, Vaarmann A, Choubey V, Safiulina D, Liiv J, Kuum M, Kaasik A (2009) PGC-1{alpha} and PGC-1{beta} regulate mitochondrial density in neurons. J Biol Chem 284, 21379-21385. 22

[16] Fleige S, Pfaffl MW (2006) RNA integrity and the effect on the real-time qRT- PCR performance. Mol Aspects Med 27, 126-139. [17] Keeney PM, Bennett JP, Jr. (2010) ALS spinal neurons show varied and reduced mtDNA gene copy numbers and increased mtDNA gene deletions. Mol Neurodegener 5, 21. [18] Keeney PM, Quigley CK, Dunham LD, Papageorge CM, Iyer S, Thomas RR, Schwarz KM, Trimmer PA, Khan SM, Portell FR, Bergquist KE, Bennett JP, Jr. (2009) Mitochondrial gene therapy augments mitochondrial physiology in a Parkinson's disease cell model. Hum Gene Ther 20, 897-907. [19] Cho SO, Ban JY, Kim JY, Jeong HY, Lee IS, Song K-S, Bae K, Seong YH (2009) –35)–Induced Neurotoxicity in Cultured Neurons and Has Antidementia Activities in Mice. JournalAralia cordata of Pharmacological Protects Against Sciences Amyloid 111 ,β 22 Protein-32. (25 [20] Coskun PE, Wyrembak J, Derbereva O, Melkonian G, Doran E, Lott IT, Head E, Cotman CW, Wallace DC (2010) Systemic mitochondrial dysfunction and the etiology of Alzheimer's disease and down syndrome dementia. J Alzheimers Dis 20 Suppl 2, S293-310. [21] Hirai K, Aliev G, Nunomura A, Fujioka H, Russell RL, Atwood CS, Johnson AB, Kress Y, Vinters HV, Tabaton M, Shimohama S, Cash AD, Siedlak SL, Harris PL, Jones PK, Petersen RB, Perry G, Smith MA (2001) Mitochondrial abnormalities in Alzheimer's disease. J Neurosci 21, 3017-3023. [22] Thomas RR, Khan SM, Smigrodzki RM, Onyango IG, Dennis J, Khan OM, Portelli FR, Bennett JP, Jr. (2012) RhTFAM treatment stimulates mitochondrial oxidative metabolism and improves memory in aged mice. Aging (Albany NY) 4, 620-635. [23] Podlesniy P, Figueiro-Silva J, Llado A, Antonell A, Sanchez-Valle R, Alcolea D, Lleo A, Molinuevo JL, Serra N, Trullas R (2013) Low CSF concentration of mitochondrial DNA in preclinical Alzheimer's disease. Ann Neurol. [24] Patti ME, Butte AJ, Crunkhorn S, Cusi K, Berria R, Kashyap S, Miyazaki Y, Kohane I, Costello M, Saccone R, Landaker EJ, Goldfine AB, Mun E, DeFronzo R, Finlayson J, Kahn CR, Mandarino LJ (2003) Coordinated reduction of genes of oxidative metabolism in humans with insulin resistance and diabetes: Potential role of PGC1 and NRF1. Proc Natl Acad Sci U S A 100, 8466-8471. [25] Cassano P, Sciancalepore AG, Pesce V, Fluck M, Hoppeler H, Calvani M, Mosconi L, Cantatore P, Gadaleta MN (2006) Acetyl-L-carnitine feeding to unloaded rats triggers in soleus muscle the coordinated expression of genes involved in mitochondrial biogenesis. Biochim Biophys Acta 1757, 1421- 1428.

Figure Legends:

Figure 1. Mitochondrial DNA copy numbers in genomic DNA isolated from groups of 20 pyramidal neurons, 20 GFAP(+) astrocytes or 15 dentate gyrus granule cells from AD or CTL hippocampal sections. Shown are histogram distributions of mtDNA copy numbers (average of gene copy number of 12S rRNA, ND2, COX III and ND4) expressed as a % of mean CTL mtDNA copy number levels. Approximately 4 caps of 15-20 cells each were isolated/case.

Figure 2. Relationships among mtDNA copy numbers in genomic DNA isolated

from groups of 20 hippocampal pyramidal neurons isolated from AD or CTL and

expression of NRF2, NRF1, TFAM and ERRα. cDNA in whole hippocampal

tissue samples from same tissues. Linear regression lines and 95% confidence

intervals are shown on each graph.

Figure 3. Relationships among expression levels of PGC1α and the expression

of other mitobiogenesis factors NRF 2, NRF 1, TFAM and ERRα from AD or CTL

hippocampal tissue samples. Linear regression lines and 95% confidence

intervals are shown on each graph.

Figure 4. Relationships between Aβ1-42 peptide levels in hippocampal tissue and

mtDNA copy numbers in pyramidal neurons, GFAP(+) astrocytes and dentate

gyrus granule cells from AD or CTL hippocampi. Linear regression lines and

95% confidence intervals are shown on each graph.

Figure 5. Effect of 48 hrs of treatment of human neural precursor cells with vehicle, 10µM Aβ25-35 or 10µM Aβ35-25 on mtDNA copy number. Values are +/-

SEM.

Supplemental Figure 1. Pilot studies demonstrating a linear response from

different numbers of hippocampal pyramidal neurons captured/LCM cap.

Supplemental Figure 2. Representative electropherograms demonstrating RNA

quality with RQI values 7.7 (upper) and 9,5 (lower)

Supplemental Figure 3. Mean mtDNA gene copy numbers of the 4 genes

assessed per LCM cap +/- SEM. There is much greater variability in mean of the

3-6 caps/case (adjacent same colored bars) than in the mean of the 4 genes

averaged/cap as demonstrated by the very small SEM values.

Supplemental Figure 4. Relationships among mtDNA copy numbers in genomic

DNA isolated from groups of 20 hippocampal pyramidal neurons isolated from

AD or CTL and expression of PGC1α. cDNA in whole hippocampal tissue

samples from same tissues. Linear regression lines and 95% confidence

intervals are shown on each graph.

Supplemental Figure 5. Relationships between Aβ1-42 peptide levels and

expression of NRF1, NRF2, TFAM and ERRα in AD and CTL hippocampal tissue

samples. Linear regression lines and 95% confidence intervals are shown on

each graph.

Supplemental Figure 6. Correlation between mtDNA copy number and relative

mitobiogenesis gene expression from AD and CTL cases vs age.

Supplemental Table ND2 probe [6 -FAM]CACGCAAGCAACCGCATCCATAAT[BHQ1a-Q] ND2 FP AAGCTGCCATCAAGTATTTCC ND2 RP GTAGTATTGGTTATGGTTCATTGTC

COX3 probe [5TET]CGAAGCCGCCGCCTGATACTG[BHQ1a-Q COX3 FP TTTCACTTTACATCCAAACATCAC COX3 RP CAATAGATGGAGACATACAGAAATAG

ND4 probe [AminoC6+TxRed]AGCCAGAACGCCTGAACGCAG[BHQ2a-Q] ND4 FP TGGCTATCATCACCCGATG ND4 RP GGTGTTGTGAGTGTAAATTAGTC

12SrRNA PROBE [Cy5]CGCCAGAACACTACGAGCCACAG[BHQ3a-Q] 12SrRNA FP CCTCAACAGTTAAATCAACAAAAC 12SrRNA RP CTGAGCAAGAGGTGGTGAC

POLg F P TGGTCAAACCCATTTCACTG RP AGAACACCTGGCTTTGGG

ERRa F P CTTCGCTCCTCCTCTCATC RP CTGGAGTCTGCTTGGAGTTAT

NRF1 F P TTTGTATGCCTTTGAAGAT RP AACCTGGATAAGTGAGAC

NRF2 F P GTTACAACTAGATGAAGAGACA RP ATCCACTGGTTTCTGACT

TFAM F P AATCTGTCTGACTCTGAA RP CACATCTCAATCTTCTACTT

PGC1a F P GATGTGAACGACTTGGAT RP TTGAAGGCTCATTGTTGTA

Figure 1.

Figure 2.

A. B.

P=0.8113 P=0.0138

C. D.

P=0.3416 P=0.046

E. F.

P=0.973 P=0.0214

G. H.

P=0.4745 P=0.0213

Figure 3. A. B.

P=0.759 P=0.017

C. D.

P=0.431 P=0.085

E. F.

P=0.191 P=0.23

G. H.

P=0.624 P=0.021

Figure 4. B. A.

P=-0.0038 P=0.5837

C. D.

P=0.3679 P=0.833

E. F.

P=0.95 P=0.7833

Figure 5.

Supplemental Figure 1. Sample 11 30

25 RQI = 7.7

Fluorescence10

0 18S 28S

20 30 40 50 60 TimeSample (seconds) 12 80

70 RQI = 9.5 60

30 Fluorescence 20

0 18S 28S -10 20 30 40 50 60 Time (seconds)

Supplemental Figure 2.

mtDNA copy number/LCM cap

16000

14000

12000

10000

8000

mtDNAnumber copy 6000

4000

2000

0 5 10 20 40 Number of cells/cap

Supplemental Figure 3.

mtDNA copy numbers/LCM cap

1e+8

1e+7

1e+6 mtDNA copy number copy mtDNA 1e+5

1e+4 1 2 3 4 5 6 7 8 9 10 1112 13 14 15 16 17 18 19

Supplemental Figure 4.

P=0.4

P=0.27

Supplemental Figure 5.

P=0.79 P=0.055

P=0.62 P=0.03

P=0.18

P=0.17

P=0.22 P=0.03